IndividualFits
SB2021_e2 (Spanton & Berry, 2021, Exp 2)
2 memory strength (low, high) x 2 encoding variability (low, high) within-subjects, between-blocks recognition memory design.
- A: High Strength, high Encoding variability
- B: High Strength, low Encoding variability
- C: Low strength, high Encoding variability
- D: Low Strength, low Encoding variability
Individual-level plots of data, model fits and predictions of equal-variance Gaussian SDT (EVSDT, 6 parameters), unequal-variance Gaussian SDT (UVSDT, 7 parameters) and Gumbel SDT (Gumbel, 6 paraneters).
Split by condition, and winning-model (in tabs).
Plots:
- A: Comparison by Delta AIC, cut-off at 15
- B : Parameter estimates. Note, in Gumbel EVSDT: \(\mu_o = -\mu_o\). In ExGauss: \(\mu_o\) designates the mean of the combined old-items distribution rather than the mean of Gaussian component, \(\beta_o = \frac{1}{\lambda_o}\) designates the scale of the exponential component, hence the mean of the Gaussian component is given by \(\mu_o - \beta_o\).
- C: zROC of data and model predictions (lines fitted by y = x^2). Note: scales vary by plot
- D: frequency of responses observed (bars) and predicted (symbols)
- E: residuals of predicted proportion of responses to observed data (+ means model predicts higher proportion of responses) in all response categories. Note: y-scale varies by plot
Condition A
High memory strength, High encoding variability
EVSDT wins (N = 15)
UVSDT wins (N = 4)
Gumbel wins (N = 40)
ExGauss-Norm wins (N = 5)
Condition B
High memory strength, Low encoding variability
EVSDT wins (N = 22)
UVSDT wins (N = 4)
Gumbel wins (N = 36)
ExGauss-Norm wins (N = 2)
Condition C
Low memory strength, high encoding variability
EVSDT wins (N = 16)
UVSDT wins (N = 0)
Gumbel wins (N = 40)
ExGauss-Norm wins (N = 8)
Condition D
Low memory strength, low encoding variability
EVSDT wins (N = 22)
UVSDT wins (N = 6)
Gumbel wins (N = 31)
ExGauss-Norm wins (N = 5)
